INDEX

Explanations

seek neutral advice

New Auto-Interp

Top Features by Cosine Similarity

Configuration

Prompts (Dashboard)

10,000 prompts, 128 tokens each

Dataset (Dashboard)

lmsys/lmsys-chat-1m

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

oci

-0.10

 vigil

-0.09

asu

-0.09

 undocumented

-0.09

idi

-0.09

swire

-0.09

 foster

-0.09

 caregivers

-0.09

 cooperation

-0.09

abi

-0.09

POSITIVE LOGITS

 neutral

0.23

Neutral

0.20

neutral

0.19

 Neutral

0.19

 third

0.18

 neutr

0.18

-neutral

0.17

 medi

0.16

 neutrality

0.15

 impartial

0.15

Activations Density 0.050%